Evaluation of mixing schemes in the HYbrid Coordinate Ocean Model (HYCOM) in the tropical Indian Ocean

Abstract

The performance of three different mixing schemes implemented in the HYbrid Coordinate Ocean Model (HYCOM), namely, K-Profile Parameterization (KPP), Goddard Institute of Space Sciences (GISS), and Mellor-Yamada (MY), is evaluated with respect to their simulation of upper ocean properties such as SST and mixed layer depth (MLD) in the tropical Indian Ocean. We analyzed interannual global HYCOM simulations without either data assimilation or SST relaxation for the recent period of 2012–2018. Our analysis shows that simulated SST is generally warmer by 1–2 °C than the observations and that there is little difference in SST between simulations by these different mixing schemes except in specific locations. The simulated MLD, irrespective of the choice of mixing scheme, in general, is deeper than observations in the tropical Indian Ocean, although this MLD bias varies with time and location depending on the mixing scheme choice. Furthermore, none of the mixing schemes analyzed consistently simulated the MLD with minimal error at all locations and for all year in the tropical Indian Ocean. Differences in the amount of cross-equatorial heat transport and the estimated thermal eddy diffusivity, especially in the eastern Indian Ocean are noted. A heat budget analysis signifies the importance of the vertical diffusive heat flux and points to the role of positive shortwave flux bias in determining the warm SST bias. The MLD biases in the simulations are not due to possible wind stress forcing errors. In addition, the wind stress-MLD relationship is stronger for these schemes compared to the observations. The KPP simulated MLD is slightly sensitive to the critical bulk Richardson number and changing it from its default value of 0.25 to 0.15 can marginally improve MLD simulation in the Indian Ocean.